United States
Environmental Protection
Agency
Risk Reduction
Engineering Laboratory
Cincinnati, OH 45268
Research and Development
EPA/540/SR-95/505 May 1995
wEPA Project Summary
Removal of RGBs from
Contaminated Soil Using the CF
Systems® Solvent Eixtraction
Process: A Treatability Study
Joseph Tillman, Lauren Drees, and Eric Saylor
The United States Environmental Pro-
tection Agency (EPA), through its Su-
perfund Technical Assistance Re-
sponse Team (START) and Super-fund
Innovative Technology Evaluation
(SITE) Programs, completed a treatabil-
ity study to determine the effectiveness
of solvent extraction in separating poly-
chlorinated biphenyls (PCBs) from soil
collected at the Springfield Township
Dump (STD) Superfund site near
Davisburg, Ml.
The study consisted of a total of six
runs, in which 100 Ib of dried soil was
processed for each run. Based on pre-
liminary analytical results, three 20-min
extraction cycles were chosen as the
most economical way to achieve the
project objectives. Therefore, this three-
extraction cycle condition was repeated
twice to acquire data for three runs
operated at the same condition. The
other three runs consisted of two, four,
and five 20-min extraction cycles.
The results of the study indicated
that on average approximately 98 per-
cent removal of PCBs was achieved
for the test runs using three extraction
cycles. The four- and five-extraction
cycle runs were the most effective in
reducing the concentration of PCBs in
soil, to 1.8 and 2.2 mg/kg, respec-
tively. This indicated that the number
of extraction cycles required for attain-
ing the lowest concentrations of PCBs
in product solids was greater than three
but less than or equal to five since
there was no discernible improvement
in PCB removal from four to five ex-
traction cycles. However, results from
oil and grease analysis suggest that
high&r removal efficiencies may be pos-
sible with additional extraction cycles.
Analysis of the filtered process wa-
ter collected from all six runs indicated
that PCBs were detected only in the
filtrate from the two-cycle run [1.9 mi-
crograms per liter [fig/L)]. PCBs were
not detected (<1.0 ng/L) in the filtrate
collected from the other five runs.
This Project Summary was developed
by EPA's START and SITE programs in
Cincinnati, OH to announce key find-
ings of a solvent extraction treatability
study that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back}.
Project Background
EPA conducted a treatability study on
soil collected from the STD Superfund site.
The approximately 4-acre site is located
near the town of Davisburg, Ml (Figure 1).
Between 1966 and 1968 the STD was
used for the disposal of drummed and
liquid industrial waste. Primary contami-
nants in the soil (a fine-to-coarse-grained
sand) include: arsenic, lead, and barium;
volatile organic compounds (VOCs); and
semivolatile organic compounds (SVOCs),
which include RGBs and the pesticide di-
eldrin.
Onsite incineration had been specified
in the Record of Decision for remediating
the soil at the site, but negative public
opinion toward incineration has led to the
consideration of treatment alternatives.
Based upon preliminary bench-scale test-
ing on soil samples taken from the site,
Printed on Recycled Paper
-------
SPRINGFIELD
TOWNSHIP *
DUMP
Figure 1. Location of the STD Superfund site.
the CF Systems® (CFS) solvent extrac-
tion process was believed to be such an
alternative. Therefore, a treatability study
was conducted to determine whether the
technology would be effective in treating
soils at the STD to the desired cleanup
standard.
Approximately 1,158 Ib of soil were ob-
tained directly from PCB "hot zones" at
the STD site and then screened onsite to
remove oversize material (>1/2-in. diam-
eter), which was approximately 168 Ib
(14.5 percent). Of the approximately
1,000 Ib of material screened to <1/2-in.
diameter, CFS used approximately 150
pounds to conduct a series of bench-scale
tests in order to establish basic operating
conditions for the treatability study. The
remaining volume was shipped to Hazen
Research, Inc. in Golden, CO, which is
the home base for CFS's Mobile Demon-
stration Unit (MDU).
The treatability study was conducted
using CFS's pilot-scale MDU on a batch-
mode. Liquified propane was the solvent
chosen to extract the organic compounds
from the STD soil. Figure 2 illustrates the
basic CFS process and the sample loca-
tions.
At Hazen Research Inc., the feed mate-
rial was air-dried, further screened to re-
move oversize material (>1/4-in. diameter),
and mixed to produce a homogenous test
feed. Table 1 summarizes the results of
the test soil screening for removal of over-
size material.
The contaminated soil was fed in 100-lb
batches into the extractor and thoroughly
mixed with approximately 150 Ib of sol-
vent for each cycle. Following phase sepa-
ration of the solvent and contaminants
from solids, the solvent/contaminant mix-
ture passes from the extraction system to
the solvent recovery system. Once in the
solvent recovery system, the solvent is
vaporized from the contaminant, con-
densed, and recycled back to the extrac-
tion system as fresh solvent. After all ex-
traction cycles are completed, water is
added to the extractor and mixed with the
solids to aid in removing any residual pro-
pane. Product solids are discharged as a
slurry in water and then filtered to form a
filter cake.
Test Objectives and Conditions
The primary objectives for the treatabil-
ity study were:
• Determine the effectiveness of
removing PCBs from STD soil to the
remedial action standard (RAS) of <1
mg/kg.
• Determine PCB concentrations in the
filtrate water to ensure proper
disposal.
Secondary objectives of the treatability
study included but were not limited to veri-
fying the absence of PCBs in the pilot-
scale unit prior to testing, determining re-
sidual concentrations of dieldrin in the
product solids, and determining mass bal-
ance for total materials.
The CFS pilot-scale treatability study
was conducted in two phases, which in-
cluded a total of five main process runs.
Phase I consisted of three test runs, each
consisting of a different number of extrac-
tion cycles. The first run consisted of three
20-min extraction cycles, the second run
consisted of four 20-min extraction cycles,
and the third run consisted of five 20-min
extraction cycles.
Preliminary analytical results, using hex-
ane as the extracting agent, indicated that
the primary objective of producing solids
having <1.0 mg/kg PCB concentration was
met for the three-cycle run. Therefore,
Phase II consisted of two additional test
runs using three 20-min extraction cycles
each, since this process condition was
believed to be the most economically fea-
sible in achieving the objective. It was
later determined that these preliminary re-
- -^- Process Path Used for
Full-Scale System
Only
0 Solid Sample Location
£?; Liquid Sample Location
Figure 2. CF3S Process Diagram.
-------
Table 1. Percentage Of Screened Oversized Material
Location
Springfield Township Dump
Hazen Research Inc.
Starting
Material (Ibs)
1,158
-626
Material
Screened (Ibs)
168*
26»
% Oversize
~ 14.5
-4.0
Total % oversize > 1/4nch
~18.Sc
a Using plastic crating having approximately 1/2-inch openings.
b Using an ASTM sieve having 1/4-inch openings.
c Oversize material could be treated following size reduction (i.e., pulverizing) during a full-scale
remediation.
suits underestimated the concentration of
PCBs in the treated soil. A sixth run
consisting of two 20-min extraction cycles
was added to test the limits of the pilot
unit in treating soil to the desired levels.
Table 2 summarizes the process condi-
tions for all six runs.
Analytical Results
Sampling was performed in accordance
with an EPA-approved Quality Assurance
Project Plan. The critical process streams
sampled for each of the six runs included:
feed soil; product solids (filter cake); and
filtrate water. Samples of the organic ex-
tract were taken at the end of Run 6, in
order to perform a mass balance on PCBs.
Table 3 summarizes the MDU's PCB
percent removal efficiencies for each run
and as averages of all six runs and the
three triple extraction cycle runs. Oil and
grease (O&G) analysis was also conducted
on feed and product solids for each test
run to determine propane's capability in
extracting semi- and nonvolatile organic
compounds in addition to PCBs. Results
of the O&G analyses are summarized in
Table 4.
PCBs were not detected in the filtrate
samples collected for the five main test
runs (<1.0 ng/L). However, for Run 6,
which involved only two extraction cycles,
PCBs were detected at 1.9 ng/L. The prod-
uct oil collected at the end of the entire
study (3700 g) contained a PCB concen-
tration in excess of 11,000 mg/kg. This
shows that the process was effective in
concentrating the PCBs within the product
oil fraction.
Dieldrin was not detected in the feed
soil nor product solids; therefore removal
could not be evaluated. More detail of the
analyses and measurements are dis-
cussed in the Project Report.
Mass Balance
A total materials balance was conducted
to account for all material loaded into the
unit for each run and to ensure that the
majority of the material was recovered
and did not simply remain in one or more
of the process components. Any material
loaded into the unit or exiting the unit was
examined in the total materials mass bal-
ance.
Table 5 summarizes the results of the
total materials balance for each run and
over the entire treatability study. Results
indicate that approximately 98% of input
material was accounted for in the product
streams. The mass balance closure for
PCBs was approximately 62%. The low
recovery is believed to be at least partially
attributable to the inability to drain all of
the oil from the extract product tank and
associated piping.
Quality of the Data
The data quality objectives established
for this pilot-scale treatability study were
based on project requirements and thus
designed to ensure that the data gener-
ated during the study would be of known
and acceptable quality to achieve the
project's technical objectives.
There were several quality control analy-
ses conducted to evaluate the laboratory
performance. These results are discussed
in detail in the full report. The critical
Table 2. Process Conditions For All Test Runs
Test Phase
1
II
Run No.
1
2
3
4
5
Feed
Loaded
(Ibs)
100
100
100
100
100
No. of
Extraction
Cycles
3
4
5
3
3
Mixing Time
Each Cycle
(min)
20
20
20
20
20
Mixing
Speed
Full
0/FulP
Full
Full
Full
Solvent/
Feed Ratio
(by weight
each cycle)
1.5/1
1.5/1
1.5/1
1.5/1
1.5/1
Extraction
Pressure
(psi)
Avg/Range
315/250-409
261/223-308
238/182-294
266/202-309
243/194-299
Extraction
Temp (°F)
Avg/Range
133/125-138
122/106-133
117/93-150
124/98-140
119/98-137
Added Run 6
100
20
Full
1.5/1
277/231-319 125/110-138
3 bottom. °ne °fthe f°Ur eXtraCt!°n CyCl8S' ^ miX6r WaS inoperable: however'a solvent «ow ™* established by recirculating propane from the top ofthe extractor into the
3
-------
Table 3. PCB Removal Efficiencies
Run
Number
1
2
3
4
5
6
No. of
Extraction
Cycles
3
4
5
3
3
2
Soil Feed
Concentration1
(mg/kg)
210
240
340
3102
220
220
Product Solids
Concentration1
(mg/kg)
4.9
1.8
2.2
4.&
5.8
19.0
Percent
Removal
97.7
99.3
99.4
98.7
97.4
91.4
Average3
260/250
6.3/4.9
97.6/98.0
1 The test method used was SW-846 3540/8080; Aroclor 1254 was the only PCB identified.
2 Average concentration of analyses of field duplicate samples (see Table 7).
3 Two values are given; the first is the average of all six runs and the second is the average of the three extraction
cycle runs (Runs 1,4, and 5).
Table 4. O&G Removal Efficiencies
Run
Number
1
2
3
4
5
6
No. of
Extraction
Cycles
3
4
5
3
3
2
Soil Feed
Concentration1
(mg/kg)
4480
4560
5870
5460
5140
7060
Product Solids
Concentration1
(mg/kg)
112
73
<20
133
93
279
Percent
Removal
97.5
98.4
>99.6
97.6
98.2
96.0
Average2
5430/5030
<118/113
>97.8/>97.8
1 Tho test method used was SW-846 9071.
2 Two values aro given; the first is the average of all six runs and the second is the average of the three extraction
cycte runs (Runs 1,4, and 5).
target analyte (PCB Aroclor 1254) was
spiked into both a sample of the product
solids and into a sample of the product
water to determine the accuracy and pre-
cision for these matrices. The results of
these matrix spike/matrix spike duplicate
(MS/MS D) samples (Table 6) show that
accuracy and precision were obtained for
both matrix types and that the project qual-
ity assurance (QA) goals were met.
Field duplicate samples of raw feed,
product solids, product oil, and product
water were collected and analyzed for
RGBs. Field duplicates provide a mea-
sure of precision for the combined samp-
ling and analytical processes. Table 7 pre-
sents these results, which show that the
project relative percent difference (RPD)
goal of 40 was easily met.
It should also be mentioned that PCBs
were not detected in any of the laboratory
method blanks, nor in a rinsate sample
collected from the pilot plant prior to the
study, indicating that contamination was
not a problem.
Conclusions
The analytical test data indicate that the
primary goal of producing solids having a
<1.0 mg/kg PCB concentration was not
attained by the designated test method
used. Of the five main test runs, the
closest PCB concentration to the HAS was
1.8 mg/kg in Run 2 product solids, which
was approximately a 99.3 percent removal
efficiency. The average PCB removal ef-
ficiency attained for the three-extraction
cycle test runs was 98%.
There is not enough PCB data to indi-
cate whether the additional fifth extraction
cycle conducted during Run 3 benefited
Table 5. Total Materials Balance
Run
Number
1
2
3
4
5
6
Total
Input (grams)
Feed
Soil1
45,400
45,800
45,800
45,800
45,800
45,800
274,400
Water
52,600
80,800
93,800
88,500
99,500
103,000
518,200
TOTAL
98,000
126,600
139,600
134,300
145,300
148,800
792,600
Output (grams)
Oil Extract
3,700
3,700
Slurry
71,600
147,400
1 16,200
134,800
141,200
158,800
770,000
F-1 Filter
Solids
485
485
485
640
640
640
3,380
Total3
72,090
147,900
116,700
135,400
141,800
163,100
777,000
(%) Recovery
Material
73.6
1172
83.6
101
97.6
110
98
1 Runs 2-6 Includs the addition of 454g of sand to fill void space in the extractor.
2 Solids not flushed out In Run 1 exited at the end of Run 2.
3 Totals rounded to four significant digits.
-------
Table 6. PCS Aroclor 1254 Matrix Spike/Matrix Spike Duplicate (MS/MSD) Results
Sample
Spike
Sample Cone.
MS
MS%R
MSD
MDS % R1
RPD*
Product Solids
Filtrate
4.9 mg/kg
10.0 ug/L
4. 1 mg/kg
< 1.0 ug/L
8.2 mg/kg
9.0 iig/L
84
90
9.1 mg/kg
7.8ng/l.
102
78
10
14
1 The OA objective for accuracy was a recovery of 50-150 percent.
C'-C,
% Recovery --
x 100
where,
2 The OA objective for precision was an RPD of<40.
(Maximum Value - Minimum Value)
C1 is the measured concentration in the spiked sample
C0 is the measured concentration in the unspiked sample
C, is the known concentration ofanalyte added to the sample
HPD
(Maximum Value + Minimum Value)/2
x 100
Table 7. PCS Aroclor 1254 Field Duplicate Results
Sample Matrix Sample Result
Result 2
RPD'
Feed
Product Solids
Product Oil
Filtrate
350 mg/kg
4.0 mg/kg
1 1,200 mg/kg
< 1 ug/L
260 mg/kg
3.9 mg/kg
1 1,300 mg/kg
< 1 ug/L
30
2.5
0.9
NC
' The project objective for precision was an RPD of<40.
NC = Not calculated
PCS removal beyond the four-extraction
cycles conducted during Run 2. The two
concentration values, for Run 2 (1.8 mg/
kg) and Run 3 (2.2 mg/kg), are essentially
equal since they are within the range of
field sampling and analytical error. How-
ever the O&G analytical results can be
used to supplement the interpretation of
results, with respect to organics removal
in general. As Table 4 indicates, when
the O&G data is evaluated, the five-ex-
traction cycles used for Run 3 appears to
have performed the best for overall or-
ganics removal.
The performance of the runs relative to
one another is illustrated in Figures 3 and
4. These show the removal of RGBs and
O&G, respectively, for each test run as
the decline in contaminant concentration
from starting feed to product solids as
sloped lines. Both figures show the dis-
parity in performance between test runs
for the respective parameters, which may
not be as apparent when simply looking
at percent removal values. Figure 3 clearly
shows that Runs 2, 3, and 4 came closer
to the test objective, assuming a feed
concentration equal to the average of all
runs (250 mg/kg). Their slopes essentially
parallel one another. Figure 3 also shows
that Runs 1 and 5 had an almost identical
performance and that Run 6 had the poor-
est performance. For O&G removal, Fig-
ure 4 indicates that Run 3 produced the
"cleanest" solids, while Runs 1, 2, 4, and
5 had similar performance. Again, Run 6
had the poorest performance, indicating
that greater than two extraction cycles are
required to achieve O&G removal efficien-
cies > 96%. These results suggest that
the extraction process operating condi-
tions could be further optimized to yield
higher removal efficiencies than were iden-
tified in this study.
Another important conclusion that re-
sulted from the study regarded the vol-
ume reduction of hazardous waste. Al-
though the CFS solvent extraction pro-
cess is not capable of destroying PCBs
and other contaminants present in the STD
soil (as is the case with solvent extraction
technologies in general), it is a means of
separating those contaminants from the
soil, thereby reducing the volume of haz-
ardous waste that must be treated. This in
turn reduces the cleanup costs involved.
The cumulative mass of the wet contami-
nated feed soil for all six runs of the treat-
ability study was approximately
274,000 g. The mass of the oily extract
sampled at the completion of Run 6 was
approximately 3,700 g. Therefore, the
process reduced the overall mass of the
contaminated material to 1.35% of its origi-
nal waste mass. The volume of the feed
soil [SG=1.34 g/mL and oil extract
(SG=0.87 g/mL) were approximately 204
and 4.3 L, respectively. Therefore, the
process reduced the overall volume"of the
contaminated material to 2.1% of its origi-
nal waste volume. The highly concen-
trated oil extracted from the CFS process
is either destroyed by incineration or
chemical dechlorination. ~
The full report was submitted in fulfill-
ment of Contract No. 68-CO-0048, Work
Assignment No. 0-50, by Science Appli-
cations International Corporation under the
sponsorship of the U.S. Environmental Pro-
tection Agency.
-------
D)
I
If)
CO
o
g
g
o
O
CD
400
300
200
100-
MJ Run No.
— — . Test Objective
" 3 extraction-cycle runs
—•— 4 extraction-cycle run
—— 5 extraction-cycle run
2 extraction-cycle run
12345
i 1 —
10 15
Product Solids Concentration (PCBs in mg/kg)
20
Figure 3. PCB removal trend.
7000 -
"55
E 6000
_c
CD
O 5000
|
g 4000
§
8
0 3000
•o
CD
<9
2000
1000
*»
*.
».
».
•^
•»
•.
•.
> ••
\
\ "\
V\ \
r^ \
y&% "\ **••.
IXX^S "-®
I \\ \
X \ * \ \
(3) ^2) \\
'T \ % ( 1 / '*'***
• * ^i^t. **"*
l \ VN S
1 \ A7-N ** **
1 , \ ®, \ \ .
© Run Ho.
••""•• 3 extraction-cycle runs
— «— 4 extraction-cycle run
— — 5 extraction-cycle run
• •• •• 2 extraction-cycle run
»4
•»
•.
*.
•»
*
»
»»
».
»»
».
••
•.
•.
•»
•.
30
50
100 150 200 250
Product Solids Concentration (O&G in mg/kg)
300
Figure 4. Oil & grease removal trend.
-------
-------
Joseph Tillman, Lauren Drees, and Eric Say/or are with Science Applications
International Corporation in Cincinnati, OH 45203.
Mark Meckes is the EPA Project Officer (see below).
The complete report (Order No. PB95-199030; Cost: $19.50, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
EPA/540/SR-95/505
------- |